Probing short and long‐range interactions in native collagen inside the bone matrix by BioSolids CryoProbe

Abstract

Solid‐state nuclear magnetic resonance is a promising technique to probe bone mineralization and interaction of collagen protein in the native state. However, many of the developments are hampered due to the low sensitivity of the technique. In this article, we report solid‐state nuclear magnetic resonance (NMR) experiments using the newly developed BioSolids CryoProbe™ to access its applicability for elucidating the atomic‐level structural details of collagen protein in native state inside the bone. We report here approximately a fourfold sensitivity enhancement in the natural abundance 13C spectrum compared with the room temperature conventional solid‐state NMR probe. With the advantage of sensitivity enhancement, we have been able to perform natural abundance 15N cross‐polarization magic angle spinning (CPMAS) and two‐dimensional (2D) 1H–13C heteronuclear correlation (HETCOR) experiments of native collagen within a reasonable timeframe. Due to high sensitivity, 2D 1H/13C HETCOR experiments have helped in detecting several short and long‐range interactions of native collagen assembly, thus significantly expanding the scope of the method to such challenging biomaterials.

Unraveling Water-Mediated 31P Relaxation in Bone Mineral

“…Previously, it has been suggested that the water layer is essential for the ordering and orientation of mineral platelets in bone . Moreover, water and its role in bone have been extensively studied by NMR spectroscopy and the relaxation methodology. ,, Recent advancements in ssNMR, including BioSolids CryoProbe and high-field dynamic nuclear polarization (DNP) based solid-state NMR instrumentation and methodologies, have solved the problem of sensitivity enhancement in bone-like complex materials to a great extent. − However, exact knowledge about the water-mediated changes inside the bone mineral is lacking, and further work is needed to characterize the role of water in bone mineralization. Additionally, the development of new solid-state NMR methodologies has become a major necessity to understand the relaxation mechanism and dynamics of bone minerals.…”

Efficient analysis of pharmaceutical drug substances and products using a solid-state NMR CryoProbe

“…Most recently, the cryogenically cooled MAS probe (CPMAS CryoProbe™, Bruker Biospin) has been reported for the structural studies of proteins with improved sensitivity. [41][42][43] Hassan et al demonstrated 3-4 fold sensitivity enhancement on heteronuclear-detected ssNMR experiments for the investigations of a wide variety of large and complex biological systems using the CPMAS CryoProbe. 42 Moreover, Lau et al integrated the CPMAS CryoProbe and X-ray diffraction (XRD) to study the structure of surfactant-like peptide fibrils, and more than 50 intermolecular peptide-peptide contacts were observed by 2D 13 C- 13 C ssNMR owing to the sensitivity boost.…”

CollagenTransformer: End-to-End Transformer Model to Predict Thermal Stability of Collagen Triple Helices Using an NLP Approach

“…Briefly, the transformer model adopts self-attention to process data out of order and learns the context of each element via positional encoding. This nonsequential method of training could be relevant to collagen, where short-range (sequential) and long-range (nonsequential) interactions play a role in the structure. , The transformer framework has increasingly become the model of choice for NLP-type of problems in language and science applications and has most recently been used in AlphaFold 2 to predict protein structures. , While transformer models are powerful, since they can be generalized to a variety of applications and modalities (sequence regression problems, sequence to sequence translation, such as secondary structure prediction, and other needs including field predictions , ), they can also be difficult to train and often require large amounts of data. This has been exemplified in recent developments of very large language models based on these architectures, sometimes reaching hundreds of billions of parameters. − Further, to our best knowledge, while a few very recent examples exist of the application of these transformer models to predict the structure or binding properties of some other protein systems, − they have thus far not been used to directly predict biophysical properties of proteins.…”

“…This nonsequential method of training could be relevant to collagen, where short-range (sequential) and longrange (nonsequential) interactions play a role in the structure. 44,45 The transformer framework has increasingly become the model of choice for NLP-type of problems in language and science applications and has most recently been used in AlphaFold 2 to predict protein structures. 46,47 While transformer models are powerful, since they can be generalized to a variety of applications and modalities (sequence regression problems, sequence to sequence translation, such as secondary structure prediction, and other needs including field predictions 48,49 ), they can also be difficult to train and often require large amounts of data.…”